Hacker News new | comments | ask | show | jobs | submit login
San Diego’s Smart Streetlights Yield a Firehose of Data (ieee.org)
72 points by teklaperry 30 days ago | hide | past | web | favorite | 76 comments

"The city is also getting ready to use the parking data to help people find a parking space. "

I can see the "real" conversations between the sales person and city officials. "We can tell you when and where exactly the parking violations is happening and guide enforcers directly to that car. This is not an expense, it is a revenue generating machine."

Every $1000 investment generate $xxxx ARR ticket revenue - easy math.

I don't know if this was meant as a negative, positive, or neutral outcome of this, but I see it as positive.

Public parking isn't going anywhere, so if this program can pay for itself by increasing parking revenue, while decreasing costs that it takes to generate that revenue (IE, more efficient checking of meters), then all the other civic benefits this data/program can create come "free". Such as improving fire response times, while decreasing cost of services:

>It would be easier if we could use the traffic signals and street lights to clear the road from Point A to Point B, and then have the [streetlight] system detect how the fire truck is moving through the street, restoring traffic flow behind it. That would save us money on fire stations, because we build them based on response times. Improving response times means we might not need as many stations.

The cynics will say this is will be used as excuse to remove resources for fire fighters, but if it can be done responsibly, and the data shows that outcomes will improve (and that outcomes do improve after implementing), I'm all for it.

One of best ways to increase parking revenue would be to set up a system to automate parking fees (similar to automated tolling).

I would gladly pay the $1 fee to park for 30 minutes in my downtown. However, I often skip paying because of the hassle of walking to the pay station, waiting several minutes to complete the transaction, and then putting the receipt back in my car.

If they just detected by license and sent me a bill, they could have my money!

There is a system that does not auto-bill in Europe but solves your issue: there are apps in which you tag with your location and license plate information. You pay within the app and you can even “top up” from within some apps as well.

The only downside, as you would expect, is that you have a new app or two for every city you travel to. It works pretty well though, and it’s way better than having to hunt down a kiosk.

I’m actually pretty surprised that kind of thing hasn’t caught on more here. I’m sure there are at least a few cities in the USA that do it.

They could have your money if you did what you were supposed to do?

While we're at it, let's automate speeding tickets. Free up the police for other crimes.

That's hard to do in California (for good reason): for speed limits under 65 (and outside school zones) there's a presumption that drivers know what they're doing.

So the speed limit may be 25 but if the prevailing traffic indicates that everyone thinks 40 is OK then you can't get ticketed for keeping up with traffic.

I see this issue come up all the time in Palo Alto where I live: people want the speed limit enforced on, say, Embarcadero Rd, but if the the city were to do so they'd have to do a traffic survey which would actually tell them to raise the limit on the signs.

There was a wave of sensible lawmaking in California that was sadly abandoned in the Wilson+ era and has never come back.

+ I don't mean Wilson was the cause, though he was a bad one; he got elected in part because people forgot about being sensible in the regard.

>So the speed limit may be 25 but if the prevailing traffic indicates that everyone thinks 40 is OK then you can't get ticketed for keeping up with traffic.

oh! that probably explains why i never in the almost 20 years here got tickets in such a situation, and i thought that i'm just good at spotting police cars (and slamming on the brakes each time i see a one) - here they are much easier to spot because they don't hide cheetah style in roadside bush like the traffic policemen do back in Russia.

Also in Cal they can't use radar unless they've recently done one of those surveys and set the radar speed threshold based on the criteria I mentioned. And they have to post a sign warning you of radar use. Of course there's no rule saying they have to remove those signs once the survey has expired (3 years I think...) so those radar in use warnings are all over the place and you can't tell which are actually true :-(

Also in California "speed traps" are illegal: this is when a highway runs through a town or district and the district puts up a sudden "25 MPH" sign so they can catch out of towners and charge them money.

That's wild. They could fix the roads to calm traffic, such as add more lights and bumps.

They do, though it's hard to put a bump on an arterial road.

Adding too many lights and stop signs speeds traffic up rather than slowing it. Fortunately Cal. seems still (mostly) to hew to the "listen popular opinions to decide what to focus on; listen to actual domain experts to decide what action to take".

There's been a lot of work done in this area, with predictable backlash. I think the issue is that the most dangerous speeding is the kind which occurs in neighbourhoods and by schools ("50km/h in a 30km/h zone"). But the easiest to apply automatic enforcement against is highway speeding ("130km/h in a 100km/h zone"), so people legitimately complain that the enforcement isn't being applied where the danger is.

The worst of it is that all the infrastructure is in place already for average speed enforcement on toll roads, since you know when everyone gets on and off the highway and all their license plates. But isn't the whole point of a toll road that you pay for the privilege of being out of traffic and going as fast as you can? Who wants to pay for a toll transponder only to get a bunch of speeding tickets in the mail the next month?

I approve in principle but currently the speed limits, especially on freeways, are unreasonably low (and that's the reason everyone breaks them, including cops driving in non-emergency situations). If the speed limits are brought up to be consistent with traffic actual speeds then I am all for automated speed ticketing of people who go much faster than the natural speed of traffic.

That's one possible interpretation, mine would be that people will always speed slightly above the posted limit because they know the odds of being ticketed are small (up to some very high speed, at which point many vehicles and/or drivers would fail). By setting them lower we are basically playing with two variables - high speed (which means getting people and goods places faster) and high risk (of higher energy collisions with less time to react). Pushing either to their limit doesn't seem like a good idea...

> That's one possible interpretation, mine would be that people will always speed slightly above the posted limit because they know the odds of being ticketed are small

Data doesn't back you up. In places where they have raised the speed limit, the overall flow speed does not increase correspondingly.

People drive at the speed they deem safe irrespective of the speed limit.

Then drive around America more often. Pretty much wherever you go, whatever the highway sign posts, people go 75-85. That includes places where the speed limit is 55 and where it is 80.

The federal limit of 55 mph was set by Nixon during the gasoline shortages due to OPEC supply control. Clinton repealed it, so some states went up to 70 mph. Today we might consider enforcing them to help reduce carbon consumption (and collision fatalities).

Heck, maybe it's time to go back to 55 as a federal limit. Last I checked, most vehicles on the road get dramatically better gas mileage at 55. Rather, it gets exponentially worse as speeds increase past 60ish due to wind resistance.

That data was true back then, but is it true now? This is a legitimate question. Cars have changed a lot since then. They're mostly plastic, more aerodynamic, and engines are more efficient. So is 55 still the average optimal?

That would tend to fit with my upper limit idea - most people aren't capable of handling their vehicles much faster than that (or reacting to other vehicles doing unexpected things). Of course, when I do drive I'm often stuck in heavy traffic which is far slower than that, so it's more of thought experiment. That said, I'd personally be surprised if people would risk a citation for reckless driving to go 85mph in a 55mph zone.

Indeed. I believe there is an upper limit of comfort for most people (somewhere in the 80-85 mph) and if the limits were raised to 80mph and then enforced at that using automated methods I would support it.

If we charge companies for polluting, it's reasonable to charge individuals for consuming carbon at an unnecessarily high rate. I'd support automatic fees enforced as drivers hit 65, 75, etc. No limit, per se, just trying to internalize the externalities.

Hard limits are strange for things that are continuous and probabilistic. If we could set an automatic fee equivalent to the cost of increased risks, pollution, etc., it'd be a function of speed. Someday we'll have the technology to communicate that back to the driver as they change speed and location from urban to rural, go through construction zones, etc.

Of course, setting an appropriate gasoline tax, paid at the pump, would reduce the value of those fees.

So if i have a solar charged Tesla, i can drive as fast as i can?

I suppose the fee would correspond to increased risk, so it'd be cheaper than for someone who was driving a gas guzzler at the same speed. In the ideal (dystopia or utopia?) world, the fee would also take into consideration average rates of collision by vehicle type, perhaps your age, number of hours spent on the road that year, etc.

We already have a solution deployed: gas taxes.

They're not high enough. And while there are gasoline taxes at the pump, there are various subsidies on the extraction and refining part of the supply chain. I'm not sure the taxes cover the subsidies. It's a wealth transfer from consumers (the ones buying gas) to the producers (the ones receiving subsidy)

Speed limits usually are for the worst case scenario: an old car, foggy and rainy day.

Sure, and for most people these speed limits are inappropriate. Forcing everyone into the worst common denominator is not a good idea for automated systems.

No, you should drive extra below speed limit on days with low visibility or reduced control.

As long as the change we well advertised and gave people time to adjust norms, I would be all for it.

That's what OnStar's for, right?

And still they won't be able to help law enforcement track down the delinquent who broke your car window or worse.

The delinquent who broke your car window has a pretty good chance of not paying whatever fine he's given (and arresting him when they come across him after a bench warrant is issued still costs money, even if he does eventually pay). You will almost certainly pay your parking ticket promptly and without causing the system to incur additional cost. Going after generally law abiding normal people has a much better ROI.

I hope government doesn’t turn into that. The legal system is supposed to maximize justice, not revenue.


The United States does not maximize revenue in the slightest. We routinely use the justice system to unnecessarily prosecute and incarcerate the poor at a huge expense.

That's because you are looking at the United States as a whole.

This police department ticketed this person for something and gets the revenue of the ticket. All other effects are costs for other departments. Positive on the books.

This Justice processing department gets budgeted for processing the ticket. All other effects are costs for other departments. Positive on the books.


This Court department gets budgeted for filing a warrant for this poor person's refusal to pay. The capturing and holding of this person will be a cost for other departments. Positive on the books.


This prison gets paid for holding this prisoner, positive on the books.

Every department is doing things which benefit them and the costs for that specific action don't show up negatively (even if they spawn later actions in other departments that generate costs for them). Every department is doing right by itself in the moment, and millions of dollars are lost.

I think you're looking at that from the wrong perspective. It's definitely generating revenue for someone but I think you're picking the wrong someone.

For-profit prison systems and perverse incentives.

In the state of Oregon, they can't prosecute car thieves, unless they literally catch them in the process of breaking in. (or the thief confesses to stealing the car) All they have to say, is they borrowed it from a friend, and didn't realize it was stolen. Something like this might actually be welcomed in Portland to cut down on car theft (its one of the top in the nation, since they literally walk right back out of the police station 2 hours later)


Right. It's OK to spy on everyone except the homeless and vagrants, because that might be offensive.

That will be driven by rfid enabled plates.

Many jurisdictions are already doing lpr driven ticketing. A guy drives down the street, he parks at the end and prints the tickets.

You don't even need RFID numberplates. The unique IDs broadcast by the (mandatory on new cars) tyre pressure monitoring sensors will do a pretty good job of uniquely identifying cars.

Most TPM systems work by comparing the difference in wheel speed using the ABS sensors though (lower pressure = lower diameter = higher wheel speed than the other 3 wheels = low air pressure warning).

Nice, I didn't know that!

Ah, no. They all work by having an air pressure sensor in the interior of the wheel.

The comment you're replying to refers to an indirect system: https://www.tirerack.com/tires/tiretech/techpage.jsp?techid=...

I'm not under the impression that indirect is the most common system, as every car that I have owned has used a direct system.

Many vehicles are switching to indirect measurement (this is just my impression, I haven't exactly been tracking which years/models use what system). It's certainly cheaper, you only need to write the algorithm once. Early systems mostly used sensors. I'm not sure why they used sensors to begin with since it's not like there's been a big improvement in VR sensors (the thing used to track wheel speed) enabling more accurate measurement. I suspect that the unwieldiness of big engineering organizations plays a role.

Many TPMS sensors don't transmit after the car is stopped for a meaningful period of time (in order to extend the battery life).

Why send enforcement? Get a neural net to determine who is illegally parked, and automatically send a ticket.

The comments about "optimizing for law enforcement" (or something like that) were a dead giveaway as to who's driving the way this system will be implemented.

San Diego resident here.

One of the more valuable things I would like to see from this network of sensors is heat maps of traffic issues.

For example, California rolling stops are the stuff of legend, and yet failure to stop is a significant cause of accidents and injuries. A heat map of the day of week, time of day, when these incidents are most likely could provide for selective and useful enforcement.

Alternatively, that same data could be used to indicate that stop signs should be replaced with a yield sign in a particular direction or convert the entire intersection to a roundabout to ease traffic flow.

My hope is that the data streams will be used in aggregate and not for individual enforcement. Trends are your friend.

This reminds me of something that happened to me several years ago.

I was in living in a small town and a new cop moved in. He became very diligent about enforcing the law, and I was pulled over for a rolling stop along a lonely country road. He didn’t give me a ticket (I wasn’t even sure what I had done wrong), but he gave me a stern warning and let me go. I continued on my way (mildly fuming) and came to the next stop down the same road. So I “stopped” and began taking a left. I was still thinking about this when I looked to my left, then right and began to pull out. I realized I just did another rolling stop, so I hit my brakes and this time I actually stopped.

And to my surprise a fast moving car emerged out of my right side blind spot. He was right at the crossing, and had I continued he would have smacked into me.

I was in my pickup truck, and the blind spot was caused by an aftermarket plastic detail that was attached to both windows, so it made the ‘A’ frame 3 times as wide as it should have been, and the angle of the road and speed of the vehicle perfectly matched the location of this plastic piece, effectively hiding the oncoming vehicle so I didn’t catch it when I first looked right.

I met that same cop some weeks later at a social gathering and of course went out of my way to shake his hand in gratitude.

I had nearly the same experience. Leaving later than usual from work, I had to exit through an entrance gate, which left my approach to the road at a different angle than usual. I looked left and right, then started into the road (doubt that I stopped completely). A car seemed to appear from nowhere. I had time to stop, but was embarrassed and agitated. The next day I recreated the scene and realized that the window frame on the right side had a slope that matched the slope of a section of the road, blocking my view. The speed and timing of our approaches and the angles of the road and window frame hid the approaching car from my sight. A full stop would have allowed for the approaching car to move through my blind spot.

Here is an article about an intersection that is at just the wrong angle, and causes deaths to cyclists:


To be pedantic (and acknowledge something I think you're acknowledging), it's the failure to yield that is the cause of the collision, not the failure to stop. One can yield without stopping, and stop without yielding. This is a pet peeve of mine, as I had a commute through a neighborhood with countless 'useless' stop signs across an arterial, and yet, as I trudged through this morass for a decade, I more than once almost failed to notice a pedestrian because I was so intent on the process of the sign, the stop line, and so on.. and less on being aware of my surroundings. The laser-like focus on the stop line actually reduced my ability to 'see' what I needed to see.

The idea of collecting lots of audiovisual data all over a city without a clear plan of what to do with it seems mad, outright dystopian to me.

The article describe how the city suddenly has much more data than what they needed or wanted and that they have to start training their people in data science now. There's no mention of privacy at all. I'd see having too much data ("But it’s like we asked for a cold drink of water and got shot in the face with a firehose") more of a liability and danger for the people of the city than a great opportunity.

Public projects should approach problems from the reverse. Figure out what problem you need to solve, then define what the minimal amount of data is that you need to capture. Make a plan how to guarantee privacy and anonymity, and if you have all that, go and put out some sensors. Not the other way round.

This is going to happen sooner or later. The efficiency gains and extra data to help the justice system come to accurate judgments is far too large. People will ultimately want it. Or cities that do it will ultimately become so well run that they will draw talent and capital from cities that refrain. Cameras and microphones everywhere will win out.

What we need to to is change the laws that these systems will enforce. Laws that were once designed to punish people harshly because they’re infrequently applied (like speeding) cannot exist once the enforcement mechanism catches it all the time.

> The efficiency gains and extra data to help the justice system come to accurate judgments is far too large

That's an admirable way of looking at the US justice system, but not quite how it really works. The system is adversarial, and "the truth" is a narrative advocated by one of the combatants.

Given enough data, a DA could choose which facts fit their narrative and present a very convincing "truth" to a jury. Unless defendants have access to the entire dataset and can select their own datapoints to refute that narrative, they're fighting at an extreme disadvantage.

Picture a scenario where a person gets murdered, and you're on trial because the only evidence the police have is a smart streetlight log of your car's whereabouts that show you parked in front of a mob-owned business and later in front of the victims house. Your truth is a simple dinner and date with the victims neighbor. The DA's truth is "backed by data".

Yes, this sounds far-fetched and may be an oversimplification, but it's important to remember that the justice system isn't programming and access to more general data doesn't necessarily level the playing field.

That's an edgy way of looking at the justice system, but not quite how it really works.

Discovery explains away most of your hypothetical.

Why is your account only 30 minutes old? You're clearly an old user. Why the throwaway account?

> Why is your account only 30 minutes old? You're clearly an old user. Why the throwaway account?

Why the combative tone, and how are you so certain that Legical is an 'old user' with a throwaway?

He understands the forum's quoting conventions. He referenced programming. He responded quickly, and got several up votes quickly but few afterwards (i.e. it looks like he extinguished his alt login voting power). Lazy username. He's writing extremist views with no justification designed to undermine confidence in the US. It just hits a little close to what I would expect from a non genuine member.

I'm not seeing anything in your explanation that couldn't be explained away as someone who had been lurking for a while and happened to decide to write a comment for the first time around the time the post fell off the front page. Besides, from the guidelines:

> Please don't impute astroturfing or shillage. That degrades discussion and is usually mistaken. If you're worried about it, email us and we'll look at the data.

On the topic of "extremist views with no justification," there are a great many people in the US who are concerned about the imbalance of power in the criminal justice system. Handwaving these concerns away with, "it's not a problem, just hire a lawyer who can go through the pile of data during discovery" doesn't strike me as an effective argument.

I strongly disagree with this somewhat defeatist view. Every privacy violation can and should be fought vigorously. It's not set in stone that we will ourselves in total surveillance in a few years time.

> People will ultimately want it

With every big data breach coming to light, more and more people get sensitized about the issues of privacy and needless data collection. People come to see privacy as a quality-of-life issue, and cities that are seen as acting too intrusive could actually be avoided.

Just having more raw data does not guarantee that a city will run better in any fashion. The task in the article about clearing the road for firefighters is a good example. The ability to set traffic lights and having a feedback through the GPS from the fire trucks is enough to achieve this task. There's nothing gained by putting a camera on the corner of every street that also watches the fire truck go by.

> With every big data breach coming to light, more and more people get sensitized about the issues of privacy and needless data collection.

Citation needed? That's not my experience at all. I find people getting more comfortable with their image and personal information basically being in the public domain--especially the younger people who have always lived in a social media driven world.

We already live much closer to this world that not, anyway. Security cameras are ubiquitous. And every Nest doorbell we add brings us closer to making impossible to walk freely without being monitored. Most people I know will trade safety and security of giving law enforcement access to this data simply to get the benefits of not having their Amazon packages stolen.

I admit there's a chance that it doesn't happen, but powerful things tend to win in a battle with ethical things. And this is very powerful. So if I had to dedicate resources to either stopping a powerful thing or fixing laws so the powerful thing isn't extra harmful, I'd pick the latter any day.

So I do live in Germany, where due to recent history, people are a tad more concerned with privacy than the residents of other countries. My statement was about observations in Germany: there was a doxxing/data leak of politicians data over christmas which was in the news for about a week. Today the collection #1 data leak made the headlines of mainstream media. Even if the conclusions often go in the wrong directions, privacy is discussed a lot.

You may argue that the situation in the US is completely different, but I'm not convinced. See for example the Apple billboard at CES. Would they have put that up if privacy wasn't interesting to their (potential) users?

You notice they put it up at CES and no where else. That's because those attendees are part of the minority population of people who care. Tim Cook talks about privacy to programmers and investors, but just has celebrities dancing to show how fun the iPhone is to the masses.

> The idea of collecting lots of data without a clear plan of what to do with it seems mad (--slightly editorialized)

And yet this is precisely what so many companies are doing now. My current employer has been talking up all the great services we can provide once we have an ocean of data about all the devices we sell. Well, we've begun condensing the vapor into an ocean and yet services (let alone a method of monetizing this data) have yet to precipitate. But let's just keep collecting because Machine Intelligence will ingest the data and save the day!

There's plenty of precedent here, there's no reasonable expectation of privacy on a public street in the US[1]. Now, that doesn't automatically mean your argument can't stand, but it becomes a quantitative one - random people snapping your picture in passing are not equivalent to the city bureau of planning feeding all that camera data to ML (unless Google / IG / etc. start picking you out of all those random photos... but I digress). Clearly the biggest experiment of this kind has been in Britain[2], no pesky 4th Amendment there, but the outcome doesn't seem to be entirely positive or negative.

1: https://en.wikipedia.org/wiki/Expectation_of_privacy#Overvie...

2: https://www.telegraph.co.uk/technology/10172298/One-surveill...

I get that there is no expectation of privacy in public, but there is a huge difference in someone maybe snapping a picture of me in the background of their Snapchat story, and the city recording my license plate on every block. One may place me at a single place and time, and will likely never see the outside of that person's photo app, the other can effectively track me wherever I go. Legal precedent really has nothing to do with how we feel about an issue.

True, but we'd need to pass new laws to prevent it, right now it's perfectly legal for the city to be doing it, that was all I meant.

I would say that one is an accident (snapping someone in the background), or rather there was no real intent. The latter is stalking.

Secure beneath the watchful eyes... only these record audio as well. Hope you didn't plan on having a conversation in private.

I remember when San Diego streetlights had radios strapped to them for the Ricochet wireless internet service. 56-128 Kbps wireless internet was a damn cool thing to see back in the late 90s/early 00s.


Oh man, we had a Ricochet model glued to the screen of our on-call laptop in Seattle in 99-2000.

Did anyone see any mention of the networking tech. their using in each light, e.g. 4G or something else?

The only mention of privacy on that page was the site's privacy policy--the issue of privacy is more important than that.

there is a military-connected startup that is making physical devices that can be placed on vehicles to track them, in that sensor network

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact